İKİNCİ ULUSAL GEOSENTETİKLER KONFERANSI

GEOSENTETİKLER DERNEĞİ G2 İKİNCİ ULUSAL GEOSENTETİKLER KONFERANSI 16-17 KASIM 2006 BOĞAZİÇİ ÜNİVERSİTESİ NATUK BİRKAN BİNASI İSTANBUL

ÖNSÖZ Türkiye Chapter ı olduğumuz International Geosynthetics Society (IGS), geosentetik ve geosentetiklerle ilgili teknolojilerin geliştirilmesi amacını taşıyan bir meslek örgütüdür. Dünya nın 68 ülkesinden 2000 kadar kişisel üyesi ve 95 kurumsal üyesi olan IGS in 26 ülkede yöresel örgütleri, yani chapter ları bulunmaktadır. Bu ülkler alfabetik sıralama ile şöyledir: Almanya (1993), Avustralya (2002), Batı Pasifik Bölgesi (1997), Belçika (2001), Brezilya (1997), Çek Cumhuriyeti (2003), Çin (1990), Fransa (1993), Güney Afrika (1995), Güney Doğu Asya (1988), Hindistan (1988), Hollanda (1992), Ispanya (1999), Italya (1992), İngiltere (1987), Japonya (1985), Kore (1993), Kuzey Amerika (1986), Meksika (2006), Peru (2001), Portekiz (2003), Romanya (1996), Şile (2006), Tayland (2002), Yunanistan (2005). Her Chapter ın adının yanında kuruluş yılı verilmiştir. Türkiye de 2001 yılında Geosentetikler Derneği olarak IGS Chapter ını kurmuştur. Dünyada tamamen Geosentetikler konusuna ayrılmış ilk konferans 1977 yılında Paris te toplanmıştır. İkinci konferans 1982 yılında toplanmış, bundan sonra da düzenli olarak her dört yılda bir toplanmaya devam etmiştir. Son dünya konferansı 2006 yılında Japonya nın Yokohama kentinde toplanmıştır. Dört yılda bir toplanan bu dünya konferansının yanısıra bölgesel konferanslar da düzenlenmektedir. Örneğin 1987 yılında ilk defa Kuzey Amerika Geosentetik Konferansı toplanmış ve ikişer yıllık peryotlarla toplanmaya devam etmiştir. 1988 yılında ise ilk Almanca dilindeki Geosentetik Konferansı toplanmış olup bu konferans da halen ikişer yıllık periyotlarla düzenli olarak toplanmaktadır. 1996 yılında Avrupa ve 1997 yılında Asya Geosentetik Bölgesel Konferansları toplanmaya başlamış ve her ikisi de dörder yıllık periyotlarla toplanmaya devam etmektedir. Doğaldır ki pek çok ülkede ulusal konferanslar da toplanmaktadır. Ülkemizdeki Birinci Ulusal Geosentetikler Konferansı Boğaziçi Üniversitesi nde toplanmış ve büyük başarı elde etmiştir. Bu ilgi ve başarı bizi İkinci Ulusal Geosentetikler Konferansı nın toplanması konusunda cesaretlendirmiştir. İkinci Ulusal Geosentetikler Konferansı na sunulmuş olan tebliğlerin kalitesi ve davetli konuşmacıların konuşmaları da bu başarının devam edeceğini göstermektedir. Artık bu konferansın iki yılda bir düzenli olarak toplanarak devam etmesi hepimizin temennisidir. Kongreye katılan bilim adamlarına, uygulamadaki mühendislere ve disiplinlerarası çalışan diğer uygulamacılara gösterdikleri büyük ilgiden dolayı yürütme kurulu adına teşekkür eder, konferansın bilim adamları ile endüstrinin kaynaşacağı ve birbirlerinin sorun ve imkanlarını tanıyacağı bir ortam oluşturmasını dilerim. Konferansın gerçekleşmesinde büyük desteği olan Boğaziçi Üniversitesi olmak üzere çok sayıdaki kuruluşa Yürütme Kurulu adına sonsuz teşekkürlerimi sunarım. Konferansın organizasyon çalışmalarında görev alan başta organizasyon komitesi olmak üzere tüm kurullara ve İnşaat Mühendisliği Bölümü araştırma görevlilerine teşekkürü bir borç bilirim. Konferansla ilgili matbaa işlerinde büyük emekleri geçen Boğaziçi Üniversitesi Yayım İşleri müdürü ve tüm personeline yürütme kurulu adına çok teşekkür eder, konferansın bilim ve endüstri yaşamına yararlı olmasını dilerim. Prof. Dr. Erol Güler Yürütme Kurulu Başkanı

BİLİM KURULU İsmail Hakkı AKSOY (İ.T.Ü.) Togan ALPER (SAMS) Ulvi ARSLAN (D.U.) Cavit ATALAR (Y.D.Ü) Cem AVCI (B.Ü.) Tuncer EDİL (U.W.) Ufuk ERGUN (O.D.T.Ü.) Telem GÖK SADIKOĞLU (İ.T.Ü.) Mete İNCECİK (İ.T.Ü.) Engin MISIRLI (K.G.M.) Kutay ÖZAYDIN (Y.T.Ü.) Ahmet SAĞLAMER (İ.T.Ü.) Ergün TOĞROL (İ.T.Ü.) Nazlı YEŞİLLER (W.S.U.) Temel YETİMOĞLU (A.Ü.) Recep YILMAZ (E.Ü.) ORGANİZASYON KOMİTESİ Orkun AKKOL İlknur BOZBEY Ayşe EDİNÇLİLER Erol GÜLER Can KÜTÜKOĞLU Hakan ÖZÇELİK (Orient Research Müşavir Mühendisler) (İstanbul Üniversitesi) (Boğaziçi Üniversitesi) (Boğaziçi Üniversitesi) (HASSAN) (East İnşaat)

İkinci Ulusal Geosentetikler Konferansı Geosentetikler Derneği ve Boğaziçi Üniversitesi tarafından düzenlenmiştir. KATKIDA BULUNAN KURULUŞLAR Orient Research Müşavir Mühendisler. Ünsa Ambalaj San. ve Tic. A.Ş. SAMS Ltd. Şti.

İÇİNDEKİLER DAVETLİ BİLDİRİLER Geogrid ile temeli güçlendirilmiş asfalt kaplamalar: Analiz yöntemi ve faydaları Erol TUTUMLUER, Imad AL-QADI, Jayhyun KWON, Samer DESSOUKY... 1 Rock PEC Geotextile Reinforced Bridge Abutment, Tanjung Malim, Malaysia Andrew HOLGATE... 21 BİLDİRİLER Stabilize Edilmiş Kohezyonlu Zeminlerdeki PVA Geogridlerin Etkileşim Özelliklerinin Değerlendirilmesi: Kesme ve Çekme Deneyleri Taner AYDOĞMUŞ, Erol GÜLER, Herbert KLAPPERICH, Tuğba ESKİŞAR... 29 Bir Model Yüzey Şerit Temelinin Donatısız ve Donatılı Kumda Davranışları Berkan MOROĞLU, Erol ŞADOĞLU, Bayram Ali UZUNER... 39 Kazandırılmış Arazilerde Geogrid Donatı Deneyleri Cavit ATALAR, Eun Chul SHIN, Braja M. DAS... 47 Geotekstil-Zemin Arayüzey Sürtünmesinin Kesme Deneyleri ile Belirlenmesi Ayşe EDİNÇLİLER, Erol GÜLER... 59 Polistiren Köpük Daneleri - Çimento - Uçucu Kül Karışımının Geoteknik Özellikleri İsmail Hakkı AKSOY, Sevilay SEVER, M. Emre HAŞAL... 69 Donatılı Kum Zeminlere Oturan Dairesel Temellerin Sayısal Analizi A. Azim YILDIZ, Mustafa LAMAN, Murat ÖRNEK, Ahmet DEMİR... 75 Geosentetik Donatılı Şevler Kullanarak Araç Yollarında Ek Şerit İmalatı S. Togan ALPER, Şükrü TIRNAKLI, Erol GÜLER... 85 Dolgu Altına İnşa Edilen Düşey Drenlerin Nümerik Analizi A. Azim YILDIZ, Minna KARSTUNEN, Harald KRENN... 93 Yüzeysel Temeller Altında Geosentetik Donatı Etkisinin Nümerik Analizi Taner AYDOĞMUŞ, A. Azim YILDIZ, Ahmet T. ARSLAN, Murat ÖRNEK... 103 Yumuşak-Orta Katı Kil Zemin Üzerine İnşa Edilen Yüksek Dolgulara Geosentetik Donatı Uygulanması Mete İNCECİK, Müge BALKAYA, Görkem TUNCA... 117

Dayanıklı Yeşil Bitki Örtüsü ve Geosentetik Donatılı Zemin Yapılarının Yangına Karşı Korunması Andreas HEROLD, Taner AYDOĞMUŞ, H. Recep YILMAZ, Pelin AKLIK... 125 Zümrütevler Kavşağında Polimer Şeritli Donatılı Zemin Köprü Kenarayağı Uygulaması Hakan ÖZÇELİK... 137 İyileştirilmiş Lös Yapılı Siltli Zeminlerde Kazıklı Radyejeneral ve Geotekstil Donatılı Dolgu ile Temel Uygulaması: Konya Oteli Örneği Mete İNCECİK, Mustafa YILDIZ, İ. Hakkı ERKAN, Murat OLGUN... 145 Temel Zemini Prefabrik Düşey Drenli Bir Deneme Dolgusu Altında Gözlenen Oturmalar Mehmet BERİLGEN, Kutay ÖZAYDIN, Sönmez YILDIRIM... 155 Geopier Kolonları Üzerinde Yeralan Geogrid Donatılı Yükleme Platformlu Bir Deneme Dolgusunun Davranışı Mehmet M. BERİLGEN, Che Hassandi ABDULLAH, Tuncer B. EDİL... 165 Geosentetiklerin filtrasyon tasarım ilkeleri Yalçın DALGIN... 177 Kaplamasız yollarda geotekstil kullanımının yararları üzerine çeşitli uygulamalar ile maliyet fayda analizleri H. Recep YILMAZ, Tuğba ESKİŞAR,, Pelin AKLIK... 191 Donatı Geosentetiklerin Uzun Dönem Dayanımları İçin Azaltma Faktörleri, Sim Metodu ve Arrhenius Modeli Yaklaşımları Pelin AKLIK, H. Recep YILMAZ, Tuğba ESKİŞAR... 201 Effect of Geosynthetics Position in Flexible Pavements Muhannad ISMEIK... 211 Geosentetik Donatılı İstinat Duvarı (Geoduvar): Bir Şartname Taslağı Erol GÜLER... 221 Alfabetik Dizin... 229

DAVETLİ BİLDİRİLER

GEOGRID BASE REINFORCED ASPHALT PAVEMENTS: ANALYSIS APPROACH AND BENEFITS GEOGRİD İLE TEMELİ GÜÇLENDİRİLMİŞ ASFALT KAPLAMALAR: ANALİZ YÖNTEMİ VE FAYDALARI Erol TUTUMLUER 1 Imad AL-QADI 2 Jayhyun KWON 3 Samer DESSOUKY 4 ABSTRACT Current ongoing research at the University of Illinois has focused on the development of a mechanistic model for the response analysis of geogrid reinforced flexible pavements. The model utilizes the finite element approach and properly considers the nonlinear, stress dependent pavement foundation geomaterials, anisotropic behavior of the granular base/subbase materials, and the compaction and preloading induced base course residual stresses. The inclusion of geogrids in the granular base has been found to establish a stiffer layer associated with aggregate interlock around the geogrid reinforcement. To validate the developed mechanistic model and also quantify the effectiveness of geogrid base reinforcement from accelerated testing, nine instrumented full-scale flexible pavement test sections were constructed for measuring pavement response and performance to vehicular axle and environmental loadings. Preliminary analyses of pavement response to wheel loading suggested that geogrid-reinforced sections experienced less tensile strain at the bottom of asphalt concrete, less vertical pressure at the top of subgrade layer, less vertical deflection in the subgrade, and significantly less lateral deflections in the aggregate base especially in the direction of traffic. This was mainly evident after the subgrade shear failure, which resulted in the highest surface rutting in the unreinforced section. 1. INTRODUCTION Geogrids and geotextiles are commonly used to increase stability and improve performance of weak subgrade by providing subgrade restraint during construction. With the latest AASHTO move towards designing pavements using mechanistic concepts in the United States (US), there has been a need to develop mechanistic models and mechanistic based design procedures to evaluate the benefits of including geosynthetics in flexible pavements. This was 1 Associate Professor, email: tutumlue@uiuc.edu 2 Founder Professor, email: alqadi@uiuc.edu 3 Graduate Research Assistant, email: jaykwon@uiuc.edu 4 Post-doctoral Research Associate, email: dsamer@uiuc.edu Department of Civil and Environmental Engineering, University of Illinois, 205 N. Mathews, Urbana, Illinois 61801, USA 1

also indicated in a recent national survey in the US which concluded that geogrids are not widely used for unbound aggregate base course reinforcement in flexible pavement applications because: (i) there is a lack of detailed knowledge on the mechanisms by which geosynthetics provide reinforcement, (ii) cost-benefit information has not been established, and (iii) acceptable design solutions are not available (Christopher et al., 2001). Absence of suitable design methodologies limits the use of geosynthetic reinforcement in roadways by contractors and state highway agencies. Geogrid base reinforcement appears to have the greatest potential for a successful and beneficial application in low to moderate volume roads having thin asphalt surfaces. When placed in a granular base course, geogrids perform as tensile reinforcement by preventing lateral spreading of the base layer. The inclusion of the geogrid also causes development of a stiffer layer associated with the interlocking action that develops around geogrid reinforcement (Konietzky et al., 2000 and 2004; Budkowska and Yu, 2003; Perkins et al., 2004; Perkins and Svano, 2004; Perkins et al., 2005). In substantially thick unbound aggregate base courses often needed for flexible airport pavements, geogrid reinforcement can also effectively reduce the base course rutting and improve the fatigue performance of the upper asphalt concrete layer. The effectiveness of geogrid reinforcement and its benefits are often realized by extending pavement life until the anticipated fatigue or rutting type mechanistic distress related failure or by reducing base course thickness with enhanced structural performance. This paper presents an overview of the recent research efforts at the University of Illinois focused on developing validated analytical tools and procedures needed to help design geogrid base reinforced flexible pavements. A mechanistic model was developed for the response analysis of geogrid reinforced flexible pavements based on the finite element approach to properly to consider the nonlinear, stress dependent pavement foundation as well as the isotropic and anisotropic behavior of the granular base/subbase materials (Kwon et al., 2005a,b). Compaction and preloading induced base course residual stresses, which are very likely to influence geogrid-aggregate interlock and development of reinforcement mechanisms, can also be considered in the mechanistic analysis for predicting critical pavement resilient responses, i.e., stresses, strains, and deformations, which can be directly linked to pavement deterioration and failure modes such as fatigue cracking and rutting. To quantify the effectiveness of geogrid base reinforcement from full-scale testing, validate the developed mechanistic model response predictions, and develop transfer functions (or distress models) for predicting rutting and fatigue performances of geogrid-reinforced pavements, a total of nine full-scale flexible pavement test sections were recently constructed at the University of Illinois (Al-Qadi et al, 2006). The subgrade soil California Bearing Ratio (CBR) was maintained below 4% for all sections. Test sections were instrumented for measuring pavement responses to vehicular axle and environmental loadings. Test section variables examined in the full-scale testing study included hot mix asphalt (HMA) and granular base layer thicknesses and type and location of geogrid within the base course. A research overview is also given in this paper of the pavement test section layout, construction, pavement instrumentation, and preliminary results and analyses of the pavement response and performance to accelerated loading. 2

2. DEVELOPED MECHANISTIC MODEL A mechanistic response model was recently developed based on the Finite Element (FE) methodology for the analysis of geosynthetic reinforced flexible pavements (Kwon et al., 2005a,b). Figure 1 shows a typical axisymmetric FE mesh for the developed mechanistic model with geosynthetic reinforcement placed at the bottom one third of the aggregate base. Continuum elements are used to model the asphalt concrete (AC), base, and subgrade layers using the 8-noded axisymmetric solid elements. In addition, 3-noded membrane elements and the neighboring 6-node interface elements, compatible with the 8-noded axisymmetric solid elements, are used to model the geosynthetic and the soil/aggregate-geosynthetic interfaces, respectively. The layered pavement system is then analyzed by assigning linear elastic or nonlinear elastic layer properties. Incremental loading is also considered in the nonlinear analysis for a proper characterization of stress dependency of the moduli and the optional inputs of residual stresses. Base Geosynthetic AC Subgrade 8 node isoparametric element Top interface element Bottom interface element Geosynthetic element Figure 1. Finite Element Mesh for the Axisymmetric Mechanistic Model Showing the Soil/Aggregate Geosynthetic Detail (Kwon et al., 2005a,b) 2.1. Asphalt Concrete and Subgrade Soil In the pavement structure, both the top asphalt concrete (AC) layer and the subgrade soil are modeled as isotropic materials. The AC layer is modeled as linear elastic with only 2 elastic material constants, elastic or resilient modulus (M R ) and Poisson s ratio (ν), used as inputs. For the subgrade layer, the nonlinear soil behavior is considered. The resilient modulus of fine-grained subgrade soils is dependent upon the stress state. Typically, soil modulus decreases in proportion to the increasing stress levels thus exhibiting stress-softening type behavior. The constitutive relationships are primarily established between the resilient 3

modulus and the deviator stress for fine-grained subgrade soils. The bilinear or arithmetic model is the most commonly used resilient modulus model for subgrade soils (Thompson and Elliott, 1985). The bilinear soil model used in the developed mechanistic model is expressed as follows: ( ) ( ) MR = K1 + K3 K2 σd When σd K2 (1) MR = K1 K4 σd K2 When σd K2 (2) where K 1, K 2, K 3, and K 4 are material constants obtained from repeated load triaxial tests and σ d (= σ 1 σ 3 ) is the deviator stress. According to Thompson and Elliot (1985), the value of the resilient modulus at the breakpoint in the bilinear curve, K 1 or E Ri, can be used to classify fine-grained soils as being soft, medium or stiff. 2.2. Unbound Aggregate Base Resilient modulus models, such as the K-θ model by Hicks and Monismith (1971) and Uzan (1985) model, consider the effects of stress dependency for modeling the nonlinear behavior of base/subbase aggregates and are generally suitable for FE programming and practical design use. Especially, the Uzan (1985) model considers the effects of both confining and deviator stresses and therefore handles very well the modulus or stiffness increase with increasing vertical and horizontal stresses in an unbound aggregate layer. The Uzan (1985) model is expressed as follows: K2 ( θ ) ( ) M = K p σ p R 1 o d o K 3 (3) where θ = σ 1 + σ 2 + σ 3 = σ 1 + 2σ 3 = bulk stress, σ d = σ 1 σ 3 = deviator stress, p 0 is the unit pressure of 1 kpa, and K 1, K 2, and K 3 are multiple regression constants obtained from repeated load triaxial test data on granular materials. Due to its simplicity and successful performance in material constant evaluation, the Uzan model was used in the developed mechanistic model as the nonlinear characterization model for granular materials. Another feature of the mechanistic model is that unbound aggregate layers can also be modeled either as isotropic with the assignment of M R and ν or cross-anisotropic, which requires the following 5 material properties: horizontal modulus (M R h ), vertical modulus (M R v ), shear modulus (G R ), and in-plane (ν h ) and out-of-plane (ν v ) Poisson s ratios. 2.3. Geosynthetic Reinforcement A three-noded axisymmetric membrane element is used in the FE mesh to model the geosynthetic reinforcement (see Figure 1). The membrane elements are capable of resisting loads in tension but they have no resistance to bending. The membrane element strain components are axial membrane strain and hoop strain in the radial and circumferential directions in the axisymmetric displacement field. This membrane element is really is a bar element in the axisymmetric analysis plane. As the axisymmetric r-z plane, shown in Figure 1, is rotated around the pavement centerline, the geosynthetic reinforcement can be modeled as a membrane due to the nature of the axismmetric stress analysis. 4

2.4. Soil/Aggregate-Geosynthetic Interface A no-thickness, six-noded interface element is used to model the soil/aggregate geosynthetic interfaces in the developed mechanistic model (see Figure 1). The 6-noded interface element has normal and shear spring stiffnesses, k s and k n, between the membrane and soil/aggregate continuum elements. The assignment of the shear and normal stiffnesses, k s and k n, depends on the magnitudes of applied normal stress and the friction properties of the interface (Desai, 2001). Various bonding conditions were assumed and studied at the geosyntheticbase/subgrade interfaces by varying the shear stiffness of the interface elements (Kwon et al., 2005b). 2.5. Residual Stresses Due to Pavement Compaction and Trafficking Compaction and preloading/trafficking induced base course residual stresses can also be considered in the mechanistic FE analysis. In a recent study by Kwon et al. (2006), this feature of the developed mechanistic model was used to investigate the development of a stiffer layer associated with aggregate interlock around the geogrid reinforcement. The primary focus was to study increased confinement effects on improved layer moduli and reduced critical subgrade vertical strains/stresses, which contributed to geogrid tensile reinforcement mechanisms. An increase in horizontal confinement due to residual stresses resulted in significant increases in the moduli of the base and subgrade layers in the vicinity of the geogrid reinforcement. The degree of structural benefit provided by geogrid reinforcement could be successfully quantified in response analysis to show the commonly observed technical benefit of geogrids in the field (Kwon et al., 2006). 3. GEOGRID BENEFITS OBSERVED FROM RESPONSE ANALYSIS Using the developed mechanistic model, Kwon et al. (2005a,b) investigated the geogrid reinforcement mechanisms by varying the shear stiffnesses in the interface elements to specify various levels of interface bonding, i.e., perfect bonding, partial bonding with geogrid and no geogrid (unreinforced). Both linear and nonlinear analyses were also performed to include isotropic and cross-anisotropic base course characterizations. In the lack of a better understanding and improved modeling of the geogrid aggregate interlock mechanism in creating confinement around geogrid reinforcement, an assignment of high geogrid stiffness, specifically 10 times the typical geogrid modulus of approximately 5,516 MPa, based on the tensile strength tests for geogrids (ASTM D 6637-01), was needed to adequately show the field observed technical advantage of geogrid reinforcement for primarily improving the stress distribution inside the base layer and reducing vertical strain on top of the subgrade soil. Figure 2 shows the effects of varying geogrid tensile modulus on the vertical subgrade strain as indicated by Kwon et al. (2005a) from both the isotropic and anisotropic characterizations of the unbound aggregate base layer considering linear and nonlinear analyses. As the geogrid modulus increases, the predicted subgrade strains effectively decreased thus showing the benefit of using geogrid reinforcement from mechanistic pavement response analysis. Nevertheless, for the typical values of geogrid modulus (300-700 MPa based on the tensile strength tests for geogrids according to ASTM D 6637-01), the benefits were quite small to negligible. 5

1600 1400 Linear Isotropic Linear Anisotropic Vertical strain @ top of subgrade ( με ) 1200 1000 800 600 400 200 Unreinforced Nonlinear Isotropic Nonlinear Anisotropic 0 1 10 100 1000 10,000 100,000 1*10 6 1*10 7 1*10 8 Geogrid Modulus (MPa) Figure 2. Effects of Varying Geogrid Modulus on Predicted Vertical Subgrade Strain (Kwon et al., 2005a) Eiksund et al. (2002) also investigated the effects of varying geogrid stiffness on reducing pavement surface deformations and concluded that relatively high values of reinforcement properties and improved interface analysis techniques were needed to best demonstrate the experimentally observed geogrid benefits, especially if a single-cycle elastic model was used in the mechanistic pavement analysis. According to recent research by Perkins et al. (2004, 2005) and Perkins and Svano (2004), residual compressive stresses built-in the unbound aggregate layers around geogrid reinforcement during construction and vehicular loading processes should help better explain the observed field benefits of geogrid reinforcement for reducing critical pavement responses. 3.1. Investigations of Aggregate Interlock and Confinement Effects using Discrete Element Modeling Recent work by Konietzky and Keip (2005) as well as McDowell et al. (2006) succesfully applied the Discrete Element Modeling (DEM) technique by the use of a 3-dimensional (3D) particle flow code, PFC3D computer program, for investigating aggregate and geogrid interactions and modeling confinement effects and the actual physical geometry of the biaxial geogrid, aggregate particles, and the aggregate-geogrid interlock (see Figure 3). In this methodology, multiple interacting bodies undergoing large dynamic motions can be modeled by modeling the individual particles or elements and computing their motion, and the overall behavior of the assembly. Force displacement laws for different element bonding conditions and the law of motion govern the movement and contact of each element for assembly of elements. 6

Figure 3. Geogrid Pullout Test Simulations to Investigate Effects of Aggregate Interlock (Konietzky and Keip, 2005) Konietzky and Keip (2005) performed PFC3D pullout test simulations using a box with a footprint of 0.112 m by 0.1161 m and a height of 0.60 m filled with a uniformly graded CA- 11 unbound aggregate material. A biaxial geogrid was placed horizontally in the middle of the aggregate column. Two to three different sized spherical elements glued together were used to establish the individual aggregate particle sizes and thereby achieve laboratory measured friction properties in model validations. In the first stage, the consolidation process with 69 kpa surface load was performed for with and without geogrid cases. Next, the surface load was removed and the horizontal stresses near the geogrid were calculated. Then, a complete pullout test was performed with the 69-kPa surface load. The complete data sets saved at intermediate stages were used for computing the relaxation. Figure 4 shows horizontal stresses near the location of geogrid computed after consolidation with and without geogrid (constant dashed and dashed-dotted lines, respectively) cases and the development of the horizontal stresses after relaxation during the pullout, each with removed surface load. S11 corresponds to the horizontal stress in the pullout direction while S22 is for the horizontal stress perpendicular to S11. What is remarkable is the significant difference between the horizontal residual stresses after consolidation with and without 7

geogrid, in particular concerning the stresses above the geogrid (see Figure 4). While the stresses without geogrid are in the range of 2 to 2.5 kpa, with geogrid they go up to 4 kpa. Although stresses are small in magnitude, with and without geogrid stress ratio is in the order of 1.5 to 2. Below the geogrid, an increase in stresses with geogrid resulted in a similar ratio of nearly 1.5. -5000-4000 - 4380-4199 -4029-3444 Residual Stresses (Pa) - 3000-2000 - 1000-3462 After 5 pullout stages - 3185-2640 - 2643 After consolidation with geogrid - 2882-2253 After consolidation without geogrid 0 S11 S22 0 0.0005 0.0010 0.0015 0.0020 0.0025 Residual Pullout Length (m) Figure 4. Development of Horizontal Residual Stresses Above the Geogrid (Konietzky and Keip, 2005) This modeling simulation proved that small geogrid movements, for example, due to the compaction process during construction or later due to traffic loading, can lead to permanent residual stresses locked-in around the geogrid, which may be directly linked to the increased confinement and interlock achieved through the use of geogrid base reinforcement in flexible pavement systems. Konietzky et al. (2004) and McDowell et al. (2006) both indicated that a stiffened zone and consequently an area of locked-in permanent residual stresses occurred approximately 10 cm above and below the geogrid, expected to vary depending on aggregate size and geogrid type. 3.2. Response Analyses of Geogrid Reinforced Pavements Using Base Course Residual Stresses As Initial Conditions Using the mechanistic FE response model, a typical low-volume road flexible pavement section was modeled to study the effects of initial residual stresses on the predicted pavement responses (Kwon et al., 2006). The layer properties assigned to the unreinforced and geogridreinforced sections are given in Table 1. Only isotropic analysis was considered in the base employing the Uzan (1985) model and the bilinear approximation by Thompson and Elliott (1985) as the nonlinear material models for the unbound aggregate base and subgrade layers, respectively. The geogrid and the interface elements were placed at the base-subgrade S11 S22 S11 S22 8

interface. Only one set of interface stiffness properties, ks and kn, were employed (see Table 1) for both the base-geogrid and geogrid-subgrade interfaces for simplicity although a more detailed study of the interface properties, currently ongoing at the University of Illinois, has indicated that typically higher shear stresses develop at the base-geogrid interface. Table 1. Material Properties Assigned In the Finite Element Analyses (Kwon et al., 2006) Materials Thickness (mm) E (MPa) ν Material Properties Asphalt Concrete 76 2,758 0.35 Isotropic and Linear Elastic Base 254 124 0.4 K 1 (kpa) K 2 K 3 4,100 0.64 0.065 K 1 or E Ri K 2 or σ di Subgrade - 41 0.45 (kpa) (kpa) K 3 K 4 Geogrid (placed at the base-subgrade interface) Interfaces (above and below the geogrid) 1.27 0 552 or 5516 41,369 200 1,000 200 0.3 Isotropic and Linear Elastic Interface Spring Stiffnesses (kpa/m) normal: k n 2,443*10 6 shear: k s 4.1*10 6 Figure 2 shows two different horizontal residual stress distributions considered in the response analyses of unreinforced and geogrid reinforced pavement sections by Kwon et al. (2006). In Figure 5(a), a residual stress of 21 kpa was assumed to exist throughout the depth of the granular base in accordance with the field measurements of Barksdale and Alba (1993). The nature and the distribution of the locked-in horizontal residual stresses in the base course around the geogrid reinforcement are still not known. After studying effects of different residual stress distributions on pavement responses, the distribution shown in Figure 5(b) with three times higher residual stresses was chosen to represent the stiffened zone in the base course directly above the geogrid for higher confinement and interlock. 76 mm AC 76 mm AC 254 mm Locked-in Residual stress 21 kpa 203 mm Residual stress 21 kpa 51 mm 62 kpa Subgrade Subgrade Case (a) Case (b) Figure 5. Pavement Geometry and Residual Stresses Assigned Throughout the Base Layer (Kwon et al., 2006) Two different modeling approaches were taken to investigate geogrid reinforcement mechanisms with the locked-in residual stresses (Kwon et al., 2006). In the first approach, a 9

typically 10 times higher than normal geogrid modulus (5,516 MPa) was first assigned to the reinforced section to somewhat amplify the tensile reinforcement effects, as considered previously by Kwon et al. (2005a,b). In this approach, a constant residual stress of 21 kpa [case (a) in Figure 5] was also assumed in the aggregate base. In the second approach, a normal geogrid modulus of 552 kpa was assigned to the reinforced sections to study the effect of increased residual stress distribution around the geogrid in the base-subgrade interface, case (b) in Figure 5, on the predicted pavement responses. Table 2 presents the critical pavement responses predicted at the centerline of loading from analyzing both the unreinforced and reinforced flexible pavement sections. Subgrade vertical strain, ε v, has been the most commonly used critical pavement response to correlate with subgrade rutting. This is also the pavement response that shows the most benefit of using geogrid tensile reinforcement in flexible pavements with its potential link to reduced subgrade permanent deformations through the use of transfer functions in the context of mechanisticempirical pavement design. Table 2. Predicted Critical Pavement Responses (Kwon et al., 2006) Pavement Response δ v (mm) surface ε R (με) bottom of AC ε v (με) top of subgrade σ v (kpa) top of subgrade Geogrid Modulus = 5,516 MPa (% reduction from unreinforced pavement) Geogrid Modulus = 552 MPa Residual stress distribution No residual stresses Case (a) Case (b) Unreinforced Reinforced Unreinforced Reinforced Reinforced 1.08-495 1,456 57.7 1.05 (3.3%) -482 (2.5%) 1047 (28.1%) 55.7 (3.5%) 1.01-466 1351 57.8 0.98 (3.3%) -451 (3.2%) 993 (26.5%) 55.3 (4.4%) 0.99-469 1089 56.0 The benefit of the geogrid reinforcement is expressed by the percent reduction in predicted responses of the unreinforced section in Table 2. The effects of using high modulus geogrid of 5,516 MPa are clearly shown for the predicted subgrade vertical strains for with and without residual stresses. In addition, Table 2 also lists the predicted responses when Figure 5 case (a) residual stresses were considered in both unreinforced and reinforced sections. In general, the presence of constant residual stresses distributed throughout the base does not really change much the degree of structural benefit provided by the geogrid reinforcement. The geogrid reinforced pavement section was analyzed in the second approach by assigning a typical biaxial geogrid modulus of 552 MPa. The residual stresses were also assumed to exist in this reinforced pavement section and the residual stress distributions were considered according to Figure 5(b). To numerically simulate stiffening around the geogrid in the second 10